Frequently I have complained that the terms "typical and atypical" or "first generation and second generation" antipsychotics were not very helpful. When I give chalk talks to junior residents and interns about antipsychotics, this is one of the first things that I note. It is the medicines relative affinities for different receptors that appear to make the difference, not whether they were discovered prior to 1980 or not. A few weeks ago, I was lucky enough to encounter a paper that was published in April of 2023 by McCutcheon et al in Biological Psychiatry titled "Data-Driven Taxonomy for Antipsychotic Medication: A New Classification System." In this episode, I use their results to guide a discussion of variability in target effects and side effects of different groups of antipsychotics.
Please leave feedback at https://www.psydactic.com.
References and readings (when available) are posted at the end of each episode transcript, located at psydactic.buzzsprout.com. All opinions expressed in this podcast are exclusively those of the person speaking and should not be confused with the opinions of anyone else. We reserve the right to be wrong. Nothing in this podcast should be treated as individual medical advice.
Frequently I have complained that the terms "typical and atypical" or "first generation and second generation" antipsychotics were not very helpful. When I give chalk talks to junior residents and interns about antipsychotics, this is one of the first things that I note. It is the medicines relative affinities for different receptors that appear to make the difference, not whether they were discovered prior to 1980 or not. A few weeks ago, I was lucky enough to encounter a paper that was published in April of 2023 by McCutcheon et al in Biological Psychiatry titled "Data-Driven Taxonomy for Antipsychotic Medication: A New Classification System." In this episode, I use their results to guide a discussion of variability in target effects and side effects of different groups of antipsychotics.
Please leave feedback at https://www.psydactic.com.
References and readings (when available) are posted at the end of each episode transcript, located at psydactic.buzzsprout.com. All opinions expressed in this podcast are exclusively those of the person speaking and should not be confused with the opinions of anyone else. We reserve the right to be wrong. Nothing in this podcast should be treated as individual medical advice.
Welcome to PsyDactic - Residency Edition. I am Dr. O’Leary and today is Friday, December 15, 2023. For those tuning in for the first time, this is a podcast about psychiatry and neuroscience. I am a 4th year psychiatry resident who does this in his spare time as a way to structure my own study patterns so I actually accomplish something. I try to be accurate and thorough, but no one is keeping me honest but myself, so you should remain skeptical of me too. I alway point out as well that the content in this podcast should in no way be confused with the opinions of the Federal Government, Department of Defense, Defense Health Agency, or the Lakin Kansas Chamber of Commerce. It is my own opinion, full stop.
About a month ago, I decided to do an episode on antipsychotics, and this first led me down the rabbit hole of dopaminergic networks in the brain, including the mesolimbic, mesocortical, nigrostriatal, and tuberoinfundibular pathways. Next, I got distracted by the neuroscience of the extrapyramidal side effects of antipsychotics, especially how blocking D2 receptors interacts with other neurotransmitter systems in the dorsal striatum, such as serotonin, acetylcholine, and glutamate to dysregulate muscle tone and movements; and how over long periods of time certain neuronal adaptations can contribute to the development of tardive dyskinesia. See my last episode for that. My last episode was entirely too long, so I am going to keep this one shorter.
Frequently I have complained that the terms typical and atypical or first generation and second generation antipsychotics were not very helpful. When I give chalk talks to junior residents and interns about antipsychotics, this is one of the first things that I note. It is the medicines relative affinities for different receptors that appear to make the difference, not whether they were discovered prior to 1980. Also, chlorpromazine, the very first antipsychotic, resembles many atypicals at least as much as it resembles other first generation antipsychotics.
A few weeks ago, I was lucky enough to encounter a paper that was published in April of 2023 by McCutcheon et al in Biological Psychiatry titled Data-Driven Taxonomy for Antipsychotic Medication: A New Classification System. They used a machine learning technique, which is basically a form of artificial intelligence, to classify antipsychotics based on their receptor affinities and whether or not they were agonists or antagonists. Then they compared these groups to the reported symptom incidents in the literature and found that membership in a group also predicted to a reasonable degree the effects and side-effects it may cause.
I am going to use this paper as a springboard to discuss particular antipsychotics. First let me review the groups that these researchers identified. There were 4 major clusters of similarity based on 3 principal components. They called these clusters 1, 2, 3, and 4 in no meaningful order, but they also gave them some meaningful names based on the shared receptor affinities that grouped them together. The groups are (1) Muscarinic, (2) Adrenergic/Low dopaminergic, (3) Serotonergic/Dopaminergic, (4) Dopaminergic. Let’s get deeper into what this means.
Group 1, the muscarinic group, is called this because its members, in addition to doing other things, are antagonists at acetylcholine muscarinic receptors M2–M5 and/or they are M1 agonists or weak antagonists. Its members include first generation antipsychotics like chlorpromazine (the very first antipsychotic) and loxapine, as well as clozapine (the first second generation antipsychotic), along with olanzapine and quetiapine. Both olanzapine and quetiapine were developed in an effort to find an alternative to clozapine, because clozapine was very effective, but also caused agranulocytosis and ultimately death of many patients. Medications in this cluster tend to cause orthostatic hypotension, are more sedating than other medications, and with long term use have high rates of metabolic side effects like hyperlipidemia, hyperglycemia and weight gain.
Group 2, the adrenergic/low DA group, consists of aripiprazole, asenapine, brexpiprazole, cariprazine, lurasidone, and ziprasidone. The authors report this group is characterized by QUOTE “low dopaminergic antagonism and … a lack of muscarinic or serotonergic antagonism but significant adrenergic antagonism and dopamine D2 partial agonism.” UNQUOTE I expected aripiprazole, brexpiprazole, and cariprazine to group together because these are all partial agonists at D2 and 5HT1A. It was surprising to find ziprasidone, lurasidone, and asenapine with them. I should point out that lurasidone had a lot of missing data. One notable commonality of this group is that, all of the members of this group are also partial agonists at 5HT1a, which is unique and different from the other groupings. When you look at the side effect profile, Group 2 is known to cause less weight gain or no weight gain and have less effect on lipid profiles than the first group and third group. Looking at the heat maps that were included in the paper, it appears that this group has a lower side effect profile in almost every category reported. The catch for group 2 is that they were reportedly also less effective at controlling both positive and negative symptoms of psychosis than the other groups. Along with fewer side effects, you may also get less overall target effectiveness.
Group 3 is the Serotonergic/Dopaminergic group. Its members include fluphenazine, haloperidol, paliperidone and risperidone, and perphenazine. These meds grouped together because of similar and significant serotonergic and dopaminergic antagonism without significant overlap with the other groups. This appeared to be the most diverse group. I would not have expected that haloperidol would fall out in this group, based on its relatively limited serotonergic activity. Despite the serotonergic activity, this group was associated with the highest levels of dystonias along with significant parkinsonism and tardive dyskinesia and substantial rates of postural hypotension and weight gain. Even though group 3 included risperidone and paliperidone, which have a reputation for causing prolactinemia, as a whole this group fell out second in the propensity to cause prolactinemia compared with group 4 which comprises the more pure dopamine blocking agents.
And that brings me to group 4, which is called the Dopaminergic group due to relatively strong dopamine receptor antagonism combined with few adrenergic effects. This group included medications that are not that commonly used in the United States. Its members were sulpiride, amisulpride, molindone, and pimozide. This group's members appeared to be more similar to each other than the other groupings, meaning the receptor profiles were less diverse, generally because they were more focussed on dopamine antagonism. While this group appeared to have higher rates of all the movement disorders it was less likely as a whole to cause metabolic side effects, with the exception of weight gain. It was also associated with the greatest improvements in both negative and positive symptoms.
Before I continue, I don’t want to leave the listener thinking that I am advocating solely for this particular data-driven grouping of antipsychotic classifications. The taxonomies we use are only as useful as the information they communicate to us. Just because principal components were calculated by a machine learning algorithm does not mean that the data used was necessarily of good quality or that these groupings are going to give practice changing information. For example, receptor profiles that are not screened would not be included. The quality of the measurements of receptor affinity themselves may vary.
Still, it is good to do things like this because looking at data from as many perspectives as we can is helpful, especially in letting us know what has strong support and what we don’t really know yet. The authors point out that different classification systems use different standards. For example the predominant neuroscience-based nomenclature in general ignores the role of muscarinic receptors in antipsychotic classification, which makes it incompetent to inform us about it. Methods like the one that they employed here, if given a large and complete data-set are potentially capable of more complete and less biased results.
Given the data that we currently have, I am going to compare some of the groups that McCutcheon et al found. There seems to be some clear signals coming through. One of these signals is in the comparison of group 1 and group 4 in terms of metabolic side effects. Group 1, the muscarinic group, is associated with the highest rates of metabolic derangement, postural hypotension, anticholinergic and metabolic side effects. Just because this group is somewhat unique in its affinity for cholinergic-muscarinic receptors does not necessarily mean that this is the mechanism of metabolic derangement, but it is suspicious. Other classes of psychotropics that are anticholinergic also increase weight gain. Tricyclic antidepressants and especially the anticholinergic SSRIs like paroxetine are associated with significant weight gain. Some have proposed various mechanisms including the increase in consumption of sugary beverages due to dry mouth, overall appetite stimulation, reduction of satiation signaling, increased cravings for high carbohydrate foods, reduced activity due to a more sedentary lifestyle caused by the sedation, or reduction of overall metabolic rate.
Given a broad look at the data, we cannot say that all antipsychotics cause metabolic issues and weight gain. The antimuscarinic ones almost certainly will cause substantial weight gain, hyperlipidemia, and hyperglycemia. Followed by the anti-serotonergic and dopaminergic ones from group 3. The low dopamine affinity /slash/ partial agonists of D2 or 5HT1A in group 2 have the lowest chance of weight gain, but still have a significant chance of metabolic issues that is higher than the more purely anti-dopaminergic agents in group 4.
Another signal that is coming through is that antipsychotics that are more specific for dopamine receptors, those in group 4 including sulpiride, amisulpride, molindone, and pimozide, appear to have better efficacy for both positive and negative symptoms of psychosis. Let me take amisulpride as an example. It is used frequently in Europe, but in the US it is approved only for nausea. It is unlikely to ever be approved for psychosis in the US because it is a generic so no pharmaceutical company is going to waste their time with an application to the FDA. It is one of the few antipsychotics that clearly demonstrates efficacy for both the positive and negative symptoms of psychosis. Amisulpride blocks both D2 and D3 receptors and that’s it, as far as we know for dopamine. There is also some evidence that it antagonizes 5HT7 from animal studies, but that is it for the serotonin receptors. It is a very clean drug. It does not bind muscarinic, histaminergic, or noradrenergic receptors. It also prefers the presynaptic D2 and D3 receptors, which means that at lower dose it can effectively increase dopamine levels at the synapse and reduce negative symptoms. At higher doses, it will spill over and block postsynaptic D2 receptors and this likely is the mechanism by which it can control positive symptoms. The real trade-off in group 4, the pure dopamine receptor blocking group, is the higher incidence of movement disorders and tardive dyskinesia. However, it shares this with group 3, the anti-dopaminergic and serotonergic group, which also has much higher rates of metabolic side effects than group 4 and with less control of positive or negative symptoms on average.
There is not a single drug in Group 4 that is used commonly in the US. Sulpride and amisulpride are not FDA approved here. Pimozide appears to have a higher than average rate of QTc prolongation which can cause cardiac arrhythmias, so was rarely used and molindone fell out of favor for unclear reasons. There is also the effect of drug companies advocating for their new drugs and this results in excitement for relatively untested drugs that appear to solve our problems, but in the end have their own problems we did not anticipate.
The CATIE trial is a famous trial of antipsychotics in the US. It was a large NIMH funded trial that was meant to answer the question whether second generation antipsychotics were more effective than first generation antipsychotics. At that point, most people just assumed from clinical experience and propaganda that second generation antipsychotics were more effective and had fewer side effects. The CATIE study compared a half dozen antipsychotics including only one first generation antipsychotic, perphenazine. It found no difference in efficacy among the drugs with 1500 patients starting the study. The fact that this study is framed as 1) comparing first to second generation antipsychotics when it only included one first generation and 2) a meaningful comparison in the first place is beyond me. McCutcheon et al, the paper I am reporting on today, clearly demonstrates that there is no biological basis for comparing first generation and second generation antipsychotics in the first place. Also, what is probably the most important finding of the CATIE trial is that ¾ of patients discontinued treatment because regardless of which antipsychotic they took, there were significant side effects. Some people give special pleading to olanzapine as being the most effective, but this was not significant, and given the huge drop-out rate, well what was it that they were comparing in the end?
In the end, what this new McCutcheon et al paper re-demonstrates is that the split between first and second generation or typical and atypical antipsychotics is not particularly predictive of efficacy or side effects. Blanket statements like typical antipsychotics primarily block D2 and atypical antipsychotics primarily or also block 5HT2A are simply not true. A focus on their actual measured receptor affinities is more predictive, but also not likely to be sufficient to make predictions of their effects or side effects. I can’t merely say that because an antipsychotic, for example, fell into cluster 2 that it won’t cause hyperlipidemia. I still need to see the data for that antipsychotic itself. However, having the ability to actually search for associations between receptor affinities and their relationship with different syndromes can help to identify better targets for research. I may get some hate mail for saying this (and you can find a form to submit it at PsyDactic.Com), but the antipsychotics we use the most in the US appear to be used more because of propaganda and politics than from actual data.
Thank you for listening. I am Dr. O and this has been an Episode of PsyDactic - Residency Edition.
https://smiadviser.org/wp-content/uploads/2021/02/Page-62-table.pdf
https://www.ncbi.nlm.nih.gov/books/NBK554403/
https://journals.physiology.org/doi/full/10.1152/jn.00254.2006
https://www.sciencedirect.com/science/article/abs/pii/S027858461830561X?via%3Dihub
https://psychscenehub.com/psychinsights/a-simplified-guide-to-antipsychotic-medications/